The Atlas method has been applied for a long time in optimal design in general and in parallel robot design in particular. With the ability to create an optimal region with many different options at the end of the problem, it opens up the opportunity to choose solutions as further technology constraints are considered. The native Atlas method needs Atlas graphics and superposition of these Atlases to obtain the optimal region. This is a subjective operation and moving towards automated design requires the elimination of such operations. This paper introduces an improvement with the Atlas method, which is to complete the problem but not graphically. Attributes are superimposed at a point by algebraically multiplying their values. The author also clarifies the transformative relationship between the working space and the design space. With all the above improvements, the design process is highly general and easily automated because it eliminates the subjective judgments of the implementers.